[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-124181-en":3,"doc-seo-124181-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},124181,1374391975076,"Riley","https://ap-avatar.wpscdn.com/avatar/14000253ca4ec9f6853?x-image-process=image/resize,m_fixed,w_180,h_180&k=1783305029341752051",8,"Research & Report","Ontology-guided machine learning outperforms zero-shot foundation models for cardiac ultrasound text reports","Ontology-guided mapping of cardiac ultrasound narrative text is examined to improve large-scale research and quality improvement. Cardiac echocardiography reports combine structured and free text that vary across institutions, making downstream text mining difficult. The study evaluates statistical machine learning (EchoMap) and zero-shot inference using GPT to map reports into a three-level hierarchical ontology across eight datasets from 24 institutions, using clinician-scored ground truth. EchoMap achieves up to 98% validation accuracy at the first level and generalizes to external data.","UCSF  \nUC San Francisco Previously Published Works  \nTitle  \nOntology-guided machine learning outperforms zero-shot foundation models for cardiac ultrasound text reports.  \nPermalink  \n[https://escholarship.org/uc/item/6sf7t41c](https://escholarship.org/uc/item/6sf7t41c)  \nJournal  \nScientific Reports, 15(1)  \nAuthors  \nSubramaniam, Suganya  \nRizvi, Sara Ramesh, Ramya et al.  \nPublication Date  \n2025-02-14  \nDOI  \n10.1038/s41598-024-83540-y  \nPeer reviewed  \n[eScholarship.org](eScholarship.org) Powered by the California Digital Library  \nUniversity of California  \n[www. nature.com/scientificreports](www. nature.com/scientificreports)  \nOPEN  \nOntology-guided machine learning outperforms zero-shot foundation models for cardiac ultrasound text reports  \nSuganya Subramaniam1, Sara Rizvi1, Ramya Ramesh2, Vibhor Sehgal2, Brinda Gurusamy2, Hikmatullah Arif3, Jeffrey Tran4, Ritu Thamman5, Emeka CAnyanwu6, Ronald Mastouri7, G. Burkhard Mackensen3 & Rima Arnaout1􀀍  \nBig data can revolutionize research and quality improvement for cardiac ultrasound. Text reports area critical part of such analyses. Cardiac ultrasound reports include structured and free text and vary across institutions, hampering attempts to mine text for useful insights. Natural language processing (NLP) can help and includes both statistical-and large language model based techniques. We tested whether we could use NLP to map cardiac ultrasound text to a three-level hierarchical ontology. We used statistical machine learning (EchoMap) and zero-shot inference using GPT. We tested eight datasets from 24 different institutions and compared both methods against clinician-scored ground truth. Despite all adhering to clinical guidelines, institutions differed in their structured reporting. EchoMap performed best with validation accuracy of 98% for the first ontology level, 93% for first and second levels, and 79% for all three. EchoMap retained performance across external test datasets and could extrapolate to examples not included in training. EchoMap’s accuracy was comparable to zeroshot GPT at the first level of the ontology and outperformed GPT at second and third levels. We show that statistical machine learning can map text to structured ontology and may be especially useful for small, specialized text datasets.  \nKeywords Natural language processing, Machine learning, Large language models, Echocardiography report, Ontology  \nBig data has the potential to revolutionize cardiac ultrasound (echocardiography) by enabling novel research and rigorous, scalable quality improvement1. Text reports are a key component of such analyses, serving as the prime means of communication for imaging findings2 and as a source of data labels for machine learning (ML) research.  \nCurrently, cardiac ultrasound reports include both structured and free text and vary across institutions, hampering attempts to mine text for useful insights. Alternatively, mapping report text to a standardized ontology can help harmonize reports across institutions, languages, and imaging modalities3.  \nSeveral medical ontologies exist4–7 and with them, tools for entity extraction and linkage. For example, the Unified Medical Language System (UMLS)4 is supported in some NLP software packages (e.g. scispacy, pymetamap). However, UMLS is focused on describing terms found in clinical notes rather than the structure and attribute details important in cardiac ultrasound reporting, as we will demonstrate below. Radlex6, developed and maintained by the Radiological Society of North America (RSNA), is not as well supported by python but contains additional attributes, for example on patient status and study protocol, that may be useful for cardiac ultrasound as well. Furthermore, researchers have used machine learning to develop cross-lingual mappingsto Radlex8. To date, however, neither cardiac ultrasound nor radiology report text is routinely mapped to an ontology in clinical practice.  \nWith respect to cardiac ultra","cbCaiugYbIJ57ljU","https://ap.wps.com/l/cbCaiugYbIJ57ljU","pdf",2730052,1,10,"English","en",105,"# Background and motivation\n## Challenges of cardiac ultrasound text reporting\n## NLP and ontology-based harmonization\n# Related work\n## Entity extraction and rule-/pattern-based approaches\n## Existing medical ontologies and tools\n# Methods and evaluation\n## EchoMap and zero-shot GPT inference\n## Datasets and clinician-scored ground truth\n## Performance results across ontology levels","[{\"question\":\"What problem does the study address in cardiac ultrasound text reporting?\",\"answer\":\"Cardiac ultrasound reports combine structured and free text that vary across institutions, which hinders reliable text mining and standardized analysis. The study targets harmonizing report text by mapping it into a structured ontology.\"},{\"question\":\"How do EchoMap and GPT approaches differ in the proposed workflow?\",\"answer\":\"EchoMap uses statistical machine learning to map text to a three-level hierarchical ontology, while GPT provides zero-shot inference for the same mapping task. Both are compared against clinician-scored ground truth.\"},{\"question\":\"What were the main performance findings across ontology levels?\",\"answer\":\"EchoMap delivered the highest validation accuracy at the first ontology level (up to 98%), with decreasing performance at deeper levels. It retained performance on external test datasets and was reported to outperform GPT at the second and third levels.\"}]","Ontology-guided machine learning outperforms zero-shot foundation models for cardiac ultrasound text reports | PDF",1785820886,25,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"ontology-guided-machine-learning-outperforms-zero-shot-foundation-models-for-cardiac-ultrasound-text-reports","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/ontology-guided-machine-learning-outperforms-zero-shot-foundation-models-for-cardiac-ultrasound-text-reports/124181/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the study address in cardiac ultrasound text reporting?","Question",{"text":75,"@type":76},"Cardiac ultrasound reports combine structured and free text that vary across institutions, which hinders reliable text mining and standardized analysis. The study targets harmonizing report text by mapping it into a structured ontology.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How do EchoMap and GPT approaches differ in the proposed workflow?",{"text":80,"@type":76},"EchoMap uses statistical machine learning to map text to a three-level hierarchical ontology, while GPT provides zero-shot inference for the same mapping task. Both are compared against clinician-scored ground truth.",{"name":82,"@type":73,"acceptedAnswer":83},"What were the main performance findings across ontology levels?",{"text":84,"@type":76},"EchoMap delivered the highest validation accuracy at the first ontology level (up to 98%), with decreasing performance at deeper levels. It retained performance on external test datasets and was reported to outperform GPT at the second and third levels.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":21,"slug":133},"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]